Main

Deficient calcium intake in premature infants can lead to a number of medical complications such as neonatal osteopenia(1). A calcium deficit is usually incurred during the first 4-8 wk of life, when only partial enteral feedings are tolerated and adequate amounts of calcium cannot be provided because of the poor solubility of calcium salts. When rapid bone growth occurs after 6-8 wk of age, fractures are more likely in these calcium-deficient infants. After that time, when full enteral feedings can be given, the calcium deficit is very gradually corrected.

In the adult, carbohydrates have been shown to facilitate the absorption of calcium(2–4). Lactose, the primary carbohydrate in human milk and most term infant formulas, has been shown to increase calcium transport in the adult, possibly by enhancing paracellular(passive, nonsaturable) calcium absorption(5, 6). Little is known regarding the potential interaction between calcium and carbohydrate absorption in the premature infant. One study suggested that net calcium retention was similar in premature infants fed formulas containing lactose or a combination of lactose and glucose polymers(7). However, this study did not compare lactose with glucose polymers alone and did not address the question of whether either carbohydrate facilitated calcium absorption(7). It appears that the ability of carbohydrates to stimulate calcium absorption is directly related to their digestibility and the absorption of the resultant monosaccharides(4, 6). Because carbohydrate malabsorption causes malabsorption of nutrients, calcium absorption could even be impaired if poorly absorbed carbohydrates are fed(8–10). Given the potential of carbohydrates to enhance or hinder calcium absorption, knowledge regarding the influence of carbohydrates on calcium absorption in the premature infant is vital. This information would be useful in designing feeding regimens which could ameliorate the potential for premature infants to develop bone disease related to poor calcium intake.

The aims of our study were to measure directly calcium absorption in the premature infant and to determine the effect of lactose and glucose polymers(the carbohydrates used in the feeding of premature infants) on the absorption of calcium. We wanted to ascertain if there is a relationship between the rate of calcium and water absorption and/or carbohydrate absorption in the premature infant. Additionally, we sought to determine directly the extent of endogenous intestinal calcium losses. To achieve these ends, we adapted the intestinal perfusion technique so that we could directly measure, in vivo intestinal nutrient flux in the premature infant(11).

METHODS

Subjects. We studied 14 low birth weight infants who were receiving full enteral orogastric tube feedings and had no active medical problem other than feeding immaturity. Informed consent was obtained from the parents, and the study was approved by the Institutional Review Board for Human Research at our institution.

Design. We studied the absorption of calcium in the absence and presence of carbohydrate using the triple lumen perfusion method(11). The calcium concentration used approximates that of premature infant formulas. We perfused the infants with three solutions differing in the carbohydrate content (Table 1): either no carbohydrate or 70 g/L of lactose or glucose polymers (Polycose, Ross Laboratories, Columbus, OH), a carbohydrate concentration approximating that used in the feeding of premature infants. PEG 4000 was added to the solutions as a nonabsorbable marker to calculate net water and calcium absorption(11). A stable isotope of calcium (44Ca) was added to ascertain lumen to mucosal flux of calcium and thus, to determine the extent of endogenous calcium loss(12). Sodium chloride concentrations were adjusted to maintain an osmolality similar to that of premature infant formulas (approximately 345 mM). Mannitol could not be used for this purpose because it has been shown to alter calcium absorption by altering water absorption(6). On the other hand, in the range of sodium concentrations used in our solutions, water absorption is not affected(13). Similarly, carbohydrate absorption is not affected within the range of sodium concentrations used(14, 15).

Table 1 Components of the perfusion solutions
Full size table

The studies were carried out at least 1 h after the previous feeding. One feeding was withheld during the course of the perfusion study. The control solution containing calcium but no carbohydrate was infused first, followed in random order by either the lactose- or glucose polymer-containing solutions. When it coincided with routine blood drawing, a serum 1,25-dihydroxyvitamin D level was obtained at the start of the perfusion.

Perfusion technique. Calcium, water, and carbohydrate absorption were determined using the triple lumen perfusion technique(11). We modified the perfusion catheters for use in premature infants (see below). The method is based on the concentration changes of a test substance (i.e. calcium) that occur in the lumen of the intestine relative to a nonabsorbable marker such as PEG 4000(11). The rate of perfusion was 0.5 mL/min. The perfusate solution was maintained between 37-38°C. After a 45-min equilibration, three 15-min samples were collected from the distal and middle collecting ports (i.e. total of six samples). Collections from the middle and distal ports were staggered by 15 min based upon our preliminary data and previous studies(11). Thus, total perfusion time was 105 min per study. The individual samples from the middle and distal ports were analyzed separately to verify that equilibration had occurred(i.e. PEG 4000 concentration between samples were within 10% of each other). The samples were then pooled for calculations.

To place the triple lumen perfusion catheter safely in a premature infant without radiographic exposure, we constructed the perfusion catheter from a 5 Fr feeding tube and two 6 Fr feeding tubes. The 6 French feeding tubes had a pH-sensitive electrode at their tip to provide immediate pH readings (Accusite pH Enteral Feeding System, Zinetics Medical, Salt Lake City, UT). The resolution of the pH sensor is 0.1 pH.

The distal port and the perfusing port were constructed from the pH-tipped tubes. The 5 Fr feeding tube served as the middle port. The ports were spaced 10 cm apart. By monitoring pH changes, the location of the perfusion catheter could be verified immediately. The tube was considered in proper position when the perfusing port pH changed from <4.0 (gastric) to >5.5 (duodenal). We use these pH-tipped tubes clinically for infants who require duodenal feedings. When we began using the catheters clinically, we obtained x-ray confirmation of tube placement before feeding. On the basis of our previous clinical experience, the perfusing port was located in the third to fourth portion of the duodenum and the distal collecting port in the proximal jejunum.

Analyses. The collected effluent was analyzed for volume and concentrations of calcium (by atomic absorption spectrophotometry), PEG 4000(by the cold acetone precipitation method)(16), lactose(Boehringer Mannheim GMBH, Mannheim, Germany), glucose polymers (Boehringer Mannheim), glucose (COBAS FARA, Roche Diagnostic Systems, Branchburg, NJ). The coefficient of variation of these assays is 5 ± 1%. The44 Ca/43Ca ratios were measured by magnetic sector thermal ionization mass spectrometry (Finnigan MAT 261, Bremen, Germany; precision of measurement was 0.1%). Serum 1,25-dihydroxyvitamin D was determined using a RIA after HPLC (Endocrine Sciences, Calabasas Hills, CA).

Calculations. Net calcium, water, and carbohydrate flux were calculated for each infant using the previously describedequations(11): whereV1 is the infusion rate; M1,M2, and M3 are the concentrations of PEG in the perfusion solution and at the proximal and distal collecting port, respectively; EP is the rate of collection at the proximal port;V2 and V3 are the flow rates at the upper and lower test segments, respectively; S2 andS3 are the solute concentrations (e.g. calcium) at the beginning and end of the test segment, respectively. ΔS is the net absorption of the solute.

Endogenous calcium loss (secretion and/or losses from cell loss) was determined as previously described by subtracting the rate of absorption (as calculated from 44Ca absorption) from the rate of net calcium flux as calculated by the above equation(12). This presumes that the absorption of 44Ca reflects the unidirectional absorption of total calcium(12). Lactose digestion and absorption were calculated as the difference between the total amount of carbohydrate(i.e. sum of lactose, glucose, and galactose) at the middle and distal ports(11). Similarly, glucose polymer digestion and absorption were calculated as the difference between the total amount of carbohydrate (i.e. total glucose after complete hydrolysis of the glucose polymers) at the middle and distal ports(11).

Statistics. The mean rates of calcium and water absorption from the three study solutions were analyzed using ANOVA. If the ANOVA testing demonstrated a significant difference among the means, Fischer's least significant difference testing was used to test for differences among the different pairs. Linear regression analysis was used to test for relationships between calcium absorption and demographic data and feeding history and water and carbohydrate absorption. Repeated measures ANOVA was used to determine whether the order of administration of the carbohydrates affected the results. The Wilcoxon signed rank test was used to test for differences in absorption between the carbohydrates.

RESULTS

Study patients. Eighteen patients were recruited to participate in the absorption study. In one patient tube placement could not be verified via appropriate pH changes, and the perfusion was not carried out. Three other subjects were excluded due to the lack of flow from one or both of the collection ports during the study. The remaining 14 tolerated the study without difficulty and had no sequelae (Table 2). Two infants were receiving human milk and the remainder standard premature infant formula (Similac Special Care 24, Ross Laboratories). All infants absorbed an amount of fluid from the study solutions that approximated the volume that would have been received from the bolus feeding the study replaced(approximately 20 mL/kg). There was no crossover effect among the studies(i.e. the order of administration did not affect the results).

Table 2 Subject demographics and feeding history
Full size table

Calcium absorption. The mean rate of calcium absorption from the solutions differed (ANOVA p = 0.03; Fig. 1). Calcium absorption was greater from the glucose polymer solution than from both the control and lactose solutions (Fig. 1). Calcium absorption from the control solution was similar to that from the lactose-containing solution (Fig. 1).

Figure 1
figure 1

Rate of calcium absorption(mmol·min-1·cm-1) from the control (calcium alone), lactose, and glucose polymer solutions. Calcium absorption was significantly greater from the glucose polymer solution compared with that from the control (p < 0.05) and lactose solutions (p< 0.02).

Full size image

Endogenous loss as measured by 44Ca absorption was 0.00093 ± 0.00016 μmol·min-1·cm-1. Thus, as a fraction of net calcium absorption, endogenous calcium loss was small (0.90 ± 0.003%). There was no correlation between calcium absorption and: gestational age, birth weight, age (in days) at the time of the study, gestational(postmenstrual) age at the time of the study, study body weight, number of days of feeding before the study, time on parenteral nutrition, or the type of feeding used before the study.

Water absorption. Water absorption followed the pattern seen for calcium absorption (ANOVA p = 0.005; Fig. 2). Water absorption was greater from the glucose polymer solution than from both the control and the lactose solutions (Fig. 2).

Figure 2
figure 2

Rate of water absorption(μL·min-1·cm-1) from the control (calcium alone), lactose, and glucose polymer solutions. Water absorption was significantly greater from the glucose polymer solution than from the control and lactose solutions.

Full size image

Carbohydrate absorption. Carbohydrate absorption was greater from the glucose polymer solution than from the lactose solution (p= 0.045; Fig. 3). There was a significant relationship between water and carbohydrate absorption (Fig. 4). If the outlier with the greatest carbohydrate absorption is left out, the relationship is still significant (r = 0.77, p = 0.001).

Figure 3
figure 3

Rate of carbohydrate absorption(mg·min-1·cm-1) from the lactose and glucose polymer solutions. Carbohydrate absorption was greater from the glucose polymer than from the lactose solution.

Full size image
Figure 4
figure 4

Relationship between water and carbohydrate absorption. Water absorption correlated positively with carbohydrate absorption.

Full size image

Calcium versus water and carbohydrate absorption. There was a positive relationship between calcium absorption and water absorption(r = 0.55, p = 0.002; Fig. 5). If the outlier with the greatest water absorption is left out, the relationship is still significant (r = 0.42, p = 0.03). Similarly, calcium absorption was also positively related to carbohydrate absorption (r= 0.46, p = 0.014; y = -0.04 + 11.7x) (data not shown).

Figure 5
figure 5

Relationship between calcium and water absorption. The rate of calcium absorption correlated positively with the rate of water absorption.

Full size image

Vitamin D. 1,25-Dihydroxyvitamin D levels were obtained in 4 of the 14 infants. All values were within the normal range: 0.13-0.21 pmol/mL(52-87 pg/mL).

DISCUSSION

Our results demonstrate that in premature infants, glucose polymers facilitate the intestinal absorption of calcium (Fig. 1). Glucose polymers enhanced the absorption of calcium, whereas there was no enhancement of calcium absorption with lactose (Fig. 1). We also have shown for the first time that in the premature infant, glucose polymers are more rapidly absorbed than is lactose (Fig. 3). There was a positive relationship between calcium and water and carbohydrate absorption (Fig. 5).

There are limited data regarding the influence of carbohydrates on calcium absorption in the premature infant. Wirth et al.(7) studied calcium absorption by measuring fecal calcium excretion in premature infants who were fed a formula containing either 100% lactose or 50% lactose and 50% glucose polymers. Although it was not statistically significant, mean calcium absorption tended to be greater from the glucose polymer-containing formula compared with the lactose containing formula (4.4 mmol·kg-1·d-1 versus 3.9 mmol·kg-1·d-1)(7); however, they did not study a solution containing 100% glucose polymers. Because our data show that glucose polymers are better digested in premature infants than lactose (Fig. 3), a higher percentage of glucose polymers in their study may have yielded similar data to our ours. In studies in term infants by Moya et al.(17) and Zeigler and Fomon(18), calcium absorption was decreased when fed with corn starch compared with lactose. Given the data from our present study, that calcium absorption is related to the digestion and absorption of the carbohydrate (Fig. 5), it implies that in the studies by Moya et al.(17) and Zeigler and Fomon(18) lactose was better digested than cornstarch. Indeed, we have previously shown in full-term infants that a significant amount of cornstarch is not digested(19).

Studies in adults have demonstrated that carbohydrates enhance the intestinal absorption of calcium(4, 6). This effect is not only related to the digestion but also to the absorption of carbohydrates co-administered with calcium(4, 6). A recent study in adults has suggested that the enhancing effect of carbohydrates on calcium absorption is due to carbohydrate-stimulated water absorption(6).

Our results fit with these observations in that the rate of carbohydrate absorption was greater from the glucose polymer solution compared with that from the lactose solution (Fig. 3). Water flux was positively related to carbohydrate absorption (Fig. 4) and calcium absorption followed water flux (Fig. 5). Our results suggest that in the premature infant, as in the adult, the enhancement of calcium absorption by carbohydrates relates to carbohydrate-stimulated water absorption (Fig. 5)(6). The relationship between calcium and water absorption (Fig. 5) implies that glucose polymers enhanced the passive absorption of calcium(6). It remains to be determined if active transport of calcium occurs in the premature infant(20).

We used 44Ca in the calcium solution to measure the endogenous loss of calcium (presumably due to secretion or loss from epithelial cells). Stable isotopes of calcium have been used previously in infants to determine indirectly the mucosal to lumen flux of calcium by comparing concentrations of the isotope in the urine and stool(21). In older children, it was demonstrated that calcium lost from the small and large intestine accounts for about 7.2% of total fecal calcium(21). Our results suggest that in the 10-cm intestinal test segment we studied, calcium secretion and/or endogenous loss accounts for less than 1% of net calcium flux. Possibly, greater intestinal losses occur more distal to the area we studied. However, the data from i.v. infused isotope calcium may overestimate stool losses(20). Previous studies using radioisotopes of calcium in adults have suggested that endogenous calcium losses were less than 1% of net calcium flux(12).

In summary, we have shown that the triple lumen method can be safely adapted for use in small premature infants. We have demonstrated that calcium absorption is enhanced by the presence of glucose polymers but not by lactose. Glucose polymers appear to be better absorbed than lactose in premature infants. The resultant enhancement in water absorption appears to relate to the ability of glucose polymers to enhance calcium absorption. These results imply that substitution of glucose polymers for lactose in premature infant formulas may increase intestinal calcium absorption and potentially decrease the incidence of medical complications associated with deficient calcium intake. Further studies are needed to confirm this hypothesis.